Conventionally, there has been developed a technology for causing a viewer to recognize a stereoscopic image by using a flat video display screen. In the technology, stereoscopic viewing is achieved by: preparing two types of videos having parallax corresponding to the distance between both eyes of a human; and causing the right eye of the viewer to visually recognize a right-eye video, and causing the left eye of the viewer to visually recognize a left-eye video.
Specifically, there is a technology for causing the viewer to recognize a stereoscopic image by: outputting a right-eye video and a left-eye video in a time-division manner, and displaying the images alternately on a single video display screen; and controlling a pair of stereoscopic glasses worn by the viewer such that a left-eye shutter is closed when the right-eye image is being displayed, and a right-eye shutter is closed when the left-eye image is being displayed.
Such multi-parallax images with a plurality of viewpoints (the right-eye video and the left-eye video) are created by various types of methods, such as generation of the multi-parallax images based on the depth estimated from one or more input images. The input images are converted into such multi-parallax images for a stereoscopic image typically by: estimating the depth in the screen from the amount of characteristics (e.g., motions and colors) of the input images; and applying parallax to the input images depending on the depth to generate parallax images.
To generate such parallax images based on the depth estimated from an input image, if the depth values serving as depth information of right and left ends of the input image are made distant from the screen surface, the outside of the screen of the input image is referred to.
Because no pixel serving as a reference is present outside of the screen of the input image to be referred to, it is difficult to generate the parallax images properly. If the viewer views a stereoscopic video based on such parallax images, the parallax images are inconsistent with each other, and the ends of the screen appear to be broken.
In view of the background described above, it is an object of the present invention to provide a stereoscopic image converting apparatus, a stereoscopic image output apparatus, and a stereoscopic image converting method capable of preventing right and left ends of a stereoscopic video generated by shifting pixels of a base image depending on the parallax amount from being broken.